Full-stack observability platform company Coralogix Ltd. today launched AI Center, a real-time observability platform designed to provide comprehensive insights into AI performance, quality, security and governance.

AI Center seeks to assist with the growing issue where as enterprises adopt AI, they require rigorous oversight, as inadequate management can lead to unforeseen or inequitable outcomes. Coralogix argues that existing AI observability approaches fall short by focusing on performance versus other attributes that impact effective usage.

Coralogix takes on observability problems with customizable evaluators that tackle the “gray areas,” such as when AI appears to perform correctly but has issues related to its responses. Differing from other vendors, Coralogix reviews the content of the user and the AI to determine whether, for example, there is a chance that an exchange contains toxicity, the AI is hallucinating, or a bad actor is trying to breach the chatbot to steal customer data.

“AI is not just another technology layer; it’s a distinct stack with its own complexities and risks,” explained Chief Executive Ariel Assaraf. “Our AI Center delivers real-time transparency into every aspect of that stack, ensuring organizations can monitor, troubleshoot and secure their AI initiatives before minor errors become major crises.”

AI Center strengthens organizations with an AI Evaluation Engine to assess the quality, correctness, security and compliance of AI applications. The service allows users to create specialized evaluators for different AI use cases, actively scanning prompts and responses for potential risks or quality issues.

With AI Security Posture Management, AI Center offers real-time monitoring of AI agents’ security and performance. The system identifies threats such as prompt injections, data leaks and personally identifiable information exposure, providing detailed dashboards to help teams quickly detect and mitigate security risks.

Another feature, User Journey and Cost Tracking, offers full transparency into AI interactions, covering conversation histories, logins and token usage. The visibility delivers the ability to detect suspicious resource consumption, prevent cost harvesting attempts and optimize budgets without sacrificing performance.

AI Center enables businesses to track AI agent accuracy, latency spikes and potential misuse, all with Performance Metrics that ensures that underperforming agents are identified and improved before they impact the user experience, creating a more reliable and efficient AI environment.

The new release encompasses technology from Aporia Technologies Ltd., a company acquired by Coralogix in December.

“The launch of our AI Center unlocks a significant barrier faced by many AI teams — crossing the chasm from pilot to production,” commented Liran Hason, vice president of AI and Coralogix and previously CEO of Aporia. “Having a centralized place to observe and manage all your AI applications for performance, quality and security is the key missing piece to launching AI apps safely.”

Coralogix is a venture capital-backed startup that has raised approximately $238 million, including a round of $142 million in June 2022. Investors in the company include Advent International Corp., Brighton Park Capital Management LP, Revaia SAS, Red Dot Capital Partners Ltd., StageOne Ventures LP, Joule Ventures Ltd., Maor Investments Ltd. and O.G. Venture Partners.

Image: Coralogix